In academic reading, an abstract is a brief overview of scholarly research. Well-written abstracts typically summarize the purpose, methodology, findings and significance of the study, allowing potential readers to quickly grasp the main points without reading the entire paper. For language lecturers engaged in EAP111, teaching students how to write high-quality abstracts and providing constructive feedback for them have been a significant emphasis.

 

However, giving feedback on students’ abstracts can sometimes be challenging, given that language lecturers on our module usually have large class sizes and are required to teach approximately 100 students each semester. As a result, problems such as the sheer volume, time constraints, and the necessity to provide tailored advice and maintain consistency can make the feedback-giving process feel like a “mission impossible”. Fortunately, XIPU AI (Junmou), a language processing AI model based on OpenAI’s GPT model, can mitigate these problems. This article will demonstrate how the integration of XIPU AI simplifies the feedback on students’ abstracts from three aspects and provide a critical evaluation regarding the boons and banes of this technological integration.

 

 

Overall, XIPU AI can streamline the feedback-giving process by commenting on three important aspects of students’ abstracts, which are organization of ideas, grammar, and vocabulary. All of them are the key features of EAP111 marking descriptors that may affect the grades of students’ writing coursework.

 

 

The first aspect that XIPU AI can help to analyze is the organization of ideas in an abstract. It can act on our prompts and check if the abstract has included all the necessary components such as research aim, methodology, and findings, which are required by our module. Furthermore, it can assess the coherence of the abstract, and provide suggestions to improve the flow of ideas and strengthen the connections between different parts of the abstract. The following screenshot displays the commentary provided by XIPU AI regarding the organization of an abstract written by a student in EAP111. The abstract was concerned with the subject of visible light communication.

 

 

 

XIPU AI has been trained on an extensive range of texts. Consequently, it is also able to identify various types of grammatical errors in students’ abstracts. In providing feedback on the grammar of students’ abstracts, it is typical to instruct XIPU AI to focus on several specific areas, including subject-verb disagreement, incorrect verb tenses, unnatural collocations, spelling mistakes and incorrect sentence structures. These are mistakes that are commonly made by students in our module. XIPU AI can accurately identify these errors, and generate suggestions for correction and improvement. The following example illustrates how XIPU AI provided feedback on the grammar of a student’s abstract whose research aim was to evaluate fire-fighting robots.

 

 

2.3 Vocabulary

One of the key criteria to evaluate if an abstract is well-written on EAP111 is whether it has included concise and accurate vocabulary. XIPU AI, with its advanced language processing capabilities, can assist us in assessing the vocabulary used in an abstract. By examining the generated feedback, it is possible to judge whether the vocabulary used in the abstract efficiently meets the aforementioned criterion. The picture below shows how XIPU AI was instructed to focus on vocabulary complexity, academic style and collocation, which are all essential requirements of the writing coursework on EAP111. The topic of this abstract was also visible light communication.

 

 

 

Based on our experience of using XIPU AI when giving feedback on students’ abstracts, both boons and banes have been identified.

 

 

 

The integration of XIPU AI works extremely well in identifying errors, thereby enabling the optimization of efficiency and the reduction of time. For example, when focusing on the vocabulary part of a student’s abstract, it is possible to request that XIPU AI ascertain whether the collocations employed therein are idiomatic. This can significantly save time for non-native English-speaking teachers. Without this technology, it would be necessary to spend hours checking them in the collocation dictionary.

 

 

The utilization of XIPU AI in the feedback-giving process facilitates the provision of continuous support for students when consistent errors are identified in their abstracts. For instance, if XIPU AI indicates that a student is constantly struggling with subject-verb agreement in his abstract, XIPU AI could be instructed to generate relevant grammar exercises and sample sentences and incorporate them into this student’s feedback. This will help to raise his awareness of this issue, strengthen his practice and avoid making the same mistake in the future.

 

 

The utilization of XIPU AI to identify and correct errors regarding organization, grammar and vocabulary allows for the reallocation of time and expertise, thereby facilitating the provision of feedback on higher-order aspects such as logic, argumentation, and critical thinking skills. These factors are conducive to students on EAP111 achieving satisfactory grades for their writing coursework. This approach can also benefit teachers, as it allows them to dedicate more time to prioritize other tasks, such as lesson planning and research.

 

 

However, XIPU AI is not impeccable. Below are some problems that we have encountered when using it to assist with our feedback.

 

 

When using XIPU AI to assess the flow of ideas in an abstract whose research topic has not been thoroughly studied or investigated, it may encounter difficulties in comprehensively grasping the pivotal concepts. For instance, when assessing the abstracts which summarize key findings in the context of eye-tracking used for assistive applications, XIPU AI exhibited deficiencies in its functionality, resulting in the generation of erroneous suggestions.

 

 

While XIPU AI can identify errors, it seems to be highly “language-oriented” and may struggle with commenting on high-level writing skills. Despite clear instructions to comment on high-level skills such as analytical and critical thinking, the generated suggestions seem to be shallow, ambiguous and vague. This often leads to confusion and misunderstanding.

 

 

Although XIPU AI is significantly accurate and advanced, it continues to depend heavily on the prompts when being utilized in the feedback. In other words, if the prompt is vague or not sufficiently detailed, this technology is unable to provide constructive feedback on abstracts.

 

 

In conclusion, XIPU AI has demonstrated considerable potential for liberating educators from the tedious feedback process and assisting them in saving time. With the time saved, they can work on other essential tasks, such as developing innovative teaching materials, providing one-on-one support to struggling students, collaborating with colleagues on research projects, or engaging in professional development activities to upgrade their teaching skills and practice. However, there still exist some obstacles, and human intervention remains indispensable in guiding this technology. This usually entails providing specific prompts and detailed instructions, in addition to the refinement of AI-generated feedback. Furthermore, different sample student abstracts are also given to update and expand the XIPU AI's knowledge base continuously. With the implementation of these measures, it is possible to ensure that practical, personalized, and constructive feedback is generated for students on EAP111 to understand how to improve their abstracts.